Author(s)

Lianfa Zou, Jian Zhang

Corresponding Author

Lianfa Zou

ABSTRACT

Aiming at the problem of low retrieval accuracy when using content-based 3D model retrieval with a single feature, a new 3D model retrieval algorithm is proposed. The method is mainly divided into the following steps: 1) Calculate the shape distribution feature and curvature feature of the 3D models; 2) Calculating a similarity value of each model in the input model and the database according to the shape distribution feature, and using the similarity value to obtain a result set; 3) Using the curvature feature of the 3D models, the similarity between each model in the result set and the input model is calculated; 4) The search result is output according to the similarity calculation result in step 3). The method realizes global and local matching of 3D models, which can effectively improve the matching precision of 3D model retrieval, and has certain application value and reference significance.

KEYWORDS

Shape, distribution, curvature, 3D model, retrieval